Where did you get the 'ideal' 29 number from?

The expected (in statistical sense) number of rotations needed is 32 but it could be as low as 24.

Now that I think it better that means the shuffle was fixed.